If we have a case of stochastic resonance, effects might be felt way before they are expected based on their size.

Also, there's a story today about the Greenland ice sheet melting 3 times faster than expected.
Metro: Arctic melt 'could sink our cities' (January 22, 2007)

...

A melt of 80km3 was predicted for 2006 but the latest figures show 287km3 has disappeared - almost three-and-a-half times more than expected.

...

Martin Truffer, a glaciologist at the University of Alaska at Fairbanks, said: 'The general thinking until very recently was that ice sheets don't react very quickly to climate.

'But that thinking is changing right now, because we're seeing things that people have thought are impossible.'

But Richard Alley, a geosciences professor at Pennsylvania State University, insisted there was no need to panic.

'There is no consensus on how much Greenland's ice will melt and no computer model that can accurately predict the future of the ice sheet,' he said.

There was a story about the Anctarctic ice sheet a few months back in which they said that basically people had been modelling the ice sheet as a single solid block of ice, but in reality it behaves like a very viscous fluid (here, geologists studying convection in the Earth's mantle could say a few things about solid rock behaving like a fluid). An ice cube will react to temperature changes rather more slowly than compressed amorphous slush.

And the fact that there is no consensus and no correct models should be no cause for comfort.

"It's the statue, man, The Statue."